使用图像的网格块从较大的图像中检测目标来训练模型并在YOLOV8n中进行预测

Question

代码演示了上传的 yolov8 模型权重，用于预测图像中的对象。我在使用球形图像作为输入来预测对象时遇到问题，任何人都可以帮助使用滑动窗口方法来执行此代码吗？

import cv2
import numpy as np
from ultralytics import YOLO
import matplotlib.pyplot as plt
import pyvips

# Load your custom YOLOv5 model
# Replace 'path/to/your/model.pt' with the correct path to your YOLOv5 model
# (Note: This may depend on how you've implemented and saved your model)
custom_yolo_model = YOLO('D:/yolo_objectdetection/runs/detect/train26/weights/best.pt')

# Define window size (adjust this based on the size of the objects you want to detect)
window_size = (1000, 1000)

# Define window stride (adjust this based on the amount of overlap between windows)
stride = (500, 500)

# Load the large input image (1333x800 dimension image)
image = cv2.imread('D:/yolo_objectdetection/newdata/images/train/Track_C-Sphere-66.jpg')

# Get image dimensions
height, width, _ = image.shape

# Initialize empty list to store detected objects
detected_objects = []
result_list = []

# Loop through the image with the sliding window
for y in range(0, height - window_size[1], stride[1]):
    for x in range(0, width - window_size[0], stride[0]):
        # Extract the current window from the image
        window = image[y:y + window_size[1], x:x + window_size[0]]
        
        # Perform object detection on the window using your custom YOLOv5 model
        results = custom_yolo_model.predict(window, show=False, save=True, save_txt=True, conf=0.7, iou=0.9)
        
        detected_objects.append(results)
        result = results[0]
        for box in result.boxes:
          class_id = result.names[box.cls[0].item()]
          cords = box.xyxy[0].tolist()
          cords = [round(x) for x in cords]
          conf = round(box.conf[0].item(), 2)
          print("Object type:", class_id)
          print("Coordinates:", cords)
          print("Probability:", conf)
          print("---")
          if len(cords) > 0:
            font = cv2.FONT_HERSHEY_SIMPLEX
            font_scale = 0.5
            font_thickness = 1
            cv2.putText(window, class_id, (cords[0], cords[1] - 5), cv2.FONT_HERSHEY_SIMPLEX, 1, (255,0,255), 1, cv2.LINE_AA)
            cv2.rectangle(window, (cords[0], cords[1]), (cords[2], cords[3]), color=(255,255,0), thickness=2)
            plt.imshow(image)